'25 at 25': Framework for Adaptive Meta-analysis (FAME) for timely, reliable and thorough results

29 Oct 2024

In the latest in our ‘25 at 25’ series, celebrating 25 years of the MRC Clinical Trials Unit at UCL, we explore FAME – a collaborative approach to prospective meta-analysis of aggregate data, which speeds up the delivery of reliable and thorough results.

Why do we need FAME?

The vast majority of systematic reviews are planned retrospectively, once most eligible trials have completed and reported their results. However, having prior knowledge of results of included trials can influence how researchers plan and conduct a systematic review and meta-analysis.

Also, most reviews and meta-analyses and are based on aggregate data (results) that can be extracted from trial publications. This means that unpublished trials and any other results that are not published cannot be included. As well as introducing a variety of reporting biases, this limits the number, consistency and reliability of the analyses.

In response, the MRC Clinical Trials Unit at UCL pioneered the framework for adaptive meta-analysis (FAME) to reduce bias, increase timeliness and provide more thorough results.

What is FAME?

FAME is a collaborative and prospective approach to meta-analysis of aggregate data.

Being prospective, FAME limits bias in the planning and conduct of systematic reviews and meta-analyses. Collaboration with trialists allows researchers to monitor how evidence from eligible trials is accumulating and identify the earliest opportunity for reliable meta-analysis. This could be months or years ahead of all trial results being available.

Then, collecting harmonised and more detailed aggregate data from trials enables researchers to carry out more consistent and thorough analyses, whilst also reducing reporting biases.

This is all achieved via the following FAME principles:

FAME in action: prostate and other cancers

In 2015, our Meta-analysis Group identified many trials (including the MRC CTU-run STAMPEDE trial) investigating the effects of adding new treatments to standard hormone therapy for people with advanced prostate cancer, which were all due to produce results.

FAME helped the team plan early, reliable meta-analyses of the effects of these treatments. By collecting consistent and more detailed data from investigators, instead of published data only, they could also do a more thorough job.

Through this series of STOPCAP meta-analyses, the team confirmed quickly and reliably that adding docetaxel or abiraterone to hormone therapy helped people with advanced prostate cancer live longer and that zoledronic acid did not. As the team requested results for different groups of patients, they could also show that the benefits of abiraterone seemed to be less in older men and that the benefits of radiotherapy to the prostate were limited to men with only a few sites of cancer spread. To date, these results have impacted 50 national and international guidelines.

In people with early prostate cancer, several trials had been investigating whether it was better to give radiotherapy immediately after surgery to remove the prostate, or to wait until there were signs that the cancer was returning. The participants in these trials (including the MRC CTU’s RADICALS-RT trial) had better prognosis than expected. This meant it was going to take a very long time before to show the effects of these radiotherapy approaches on disease spread beyond the prostate, or on how long people lived.

Therefore, in this FAME meta-analysis, the ARTISTIC collaboration planned to assess the effects of these treatments on an earlier outcome measure - whether the cancer worsened or returned. There was no evidence that immediate radiotherapy improved this outcome, meaning many people with prostate cancer could be spared the nasty side-effects of radiotherapy until their disease worsened.

This example illustrates how FAME can also benefit included trials. With the trials taking longer to complete than expected, the plan for a meta-analysis reassured independent data monitoring committees and funders that it was worth continuing trial recruitment, follow-up and funding. It also justified a change to the primary outcome of RADICALS-RT. Additionally, by doing the meta-analysis alongside the trials analyses, the teams were able to publish them all on the same day, for bigger impact. Representatives from the included trials contributed to and co-authored the meta-analysis paper. So far, these results have been cited in 14 national and international guidelines.

An ongoing FAME meta-analysis for the Adjuvant Aspirin Trialists Collaboration (which includes the MRC CTU’s Add-Aspirin trial) will explore whether taking aspirin after having treatment for colorectal cancer lowers the risk of the cancer returning. The meta-analysis will also investigate whether any effect of aspirin varies by dose or duration, or by particular patient or tumour characteristics.

‘FAME-lite’ as an alternative: from pandemic back to prostate cancer

It may not always be feasible to use all the principles of FAME. This might be, for example, because some relevant trials are already published, or if trials are producing results very quickly. However, using even some FAME principles (FAME-lite) can still improve the timeliness, reliability and thoroughness of meta-analysis findings. This is exactly what was achieved during the COVID-19 pandemic.

Our researchers assisted the World Health Organization (WHO) REACT team in a series of collaborative, prospective meta-analyses. One examined the effects of anti-IL6 agents for patients in hospital with COVID-19. The team collected data from 27 trials from 28 countries, and 18 trialists supplied their results before publication.

The meta-analysis showed that anti-IL6 agents reduced the risk of dying within 28 days of hospital admission. Patients who received IL6 agents alongside corticosteroids did even better. These results led to a change in WHO guidelines, which were published on the same day as the meta-analysis.

Currently, our researchers are investigating the effects of hormone therapy after surgery for early prostate cancer. When the team started to address this research question, three trials had already published their results. RADICALS-HD, run by the MRC CTU, had not. The team quickly established the DADSPORT collaboration to plan to conduct the most detailed and reliable meta-analysis possible, before the RADICALS-HD results came out.

Preliminary DADSPORT results showed that six months of hormone therapy seems to reduce the risk of the cancer spreading, but hormone therapy does not appear to improve how long people live. These findings were presented alongside results from RADICALS-HD trial and full results are expected soon.

Recently, researchers showcased the principles of FAME at the International Clinical Trials Methodology Conference. Jayne Tierney presented a poster and Claire Vale gave a talk on the benefits involving trialists in systematic reviews and meta-analyses.

Further information: